Patent Application
20130092968
The present invention relates to a nitride-based semiconductor device and a method for fabricating such a device. More particularly, the present invention relates to a GaN-based semiconductor light-emitting device such as a light-emitting diode or a laser diode that operates at wavelengths over the entire visible radiation range, which covers the ultraviolet, blue, green, orange and white parts of the spectrum. Such a light-emitting device is expected to be applied to various fields of technologies including display, illumination and optical information processing in the near future. The present invention also relates to a method of making an electrode for use in such a nitride-based semiconductor device.
A nitride semiconductor including nitrogen (N) as a Group V element is a prime candidate for a material to make a short-wave light-emitting device because its bandgap is sufficiently wide. Among other things, gallium nitride-based compound semiconductors (which will be referred to herein as “GaN-based semiconductors” and which are represented by the formula AlxGayInzN (where 0≦x, y, z≦1 and x+y+z=1)) have been researched and developed particularly extensively. As a result, blue light-emitting diodes (LEDs), green LEDs, and semiconductor laser diodes made of GaN-based semiconductors have already been used in actual products (see Patent Documents Nos. 1 and 2, for example).
A GaN-based semiconductor has a wurtzite crystal structure.
In fabricating a semiconductor device using GaN-based semiconductors, a c-plane substrate, i.e., a substrate of which the principal surface is a (0001) plane, is used as a substrate on which GaN semiconductor crystals will be grown. In a c-plane, however, there is a slight shift in the c-axis direction between a Ga atom layer and a nitrogen atom layer, thus producing electrical polarization there. That is why the c-plane is also called a “polar plane”. As a result of the electrical polarization, a piezoelectric field is generated in the InGaN quantum well of the active layer in the c-axis direction. Once such a piezoelectric field has been generated in the active layer, some positional deviation occurs in the distributions of electrons and holes in the active layer due to the quantum confinement Stark effect of carriers. Consequently, the internal quantum yield decreases, thus increasing the threshold current in a semiconductor laser diode and increasing the power dissipation and decreasing the luminous efficacy in an LED. Meanwhile, as the density of injected carriers increases, the piezoelectric field is screened, thus varying the emission wavelength, too.
Thus, to overcome these problems, it has been proposed that a substrate, of which the principal surface is a non-polar plane such as a (10-10) plane that is perpendicular to the [10-10] direction and that is called an “m-plane”, be used. As used herein, “−” attached on the left-hand side of a Miller-Bravais index in the parentheses means a “bar” (a negative direction index). As shown in
In this case, the “m-plane” is a generic term that collectively refers to a family of planes including (10-10), (-1010), (1-100), (-1100), (01-10) and (0-110) planes. Also, as used herein, the “X-plane growth” means epitaxial growth that is produced perpendicularly to the X plane (where X=c or m) of a hexagonal wurtzite structure. As for the X-plane growth, the X plane will be sometimes referred to herein as a “growing plane”. Furthermore, a layer of semiconductor crystals that have been formed as a result of the X-plane growth will be sometimes referred to herein as an “X-plane semiconductor layer”.
As described above, a GaN-based semiconductor device that has been grown on an m-plane substrate would achieve far more beneficial effects than what has been grown on a c-plane substrate but still has the following drawback. Specifically, a GaN-based semiconductor device that has been grown on an m-plane substrate has higher contact resistance than what has been grown on a c-plane substrate, which constitutes a serious technical obstacle to using such a GaN-based semiconductor device that has been grown on an m-plane substrate.
Under the circumstances such as these, the present inventors carried out extensive research to overcome such a problem with the prior art that a GaN-based semiconductor device, grown on an m-plane as a non-polar plane, would have high contact resistance. As a result, we found an effective means for reducing the contact resistance.
It is therefore an object of the present invention to provide a structure and manufacturing process that will be able to reduce the contact resistance of a GaN-based semiconductor device that has been fabricated by producing a crystal growth on an m-plane substrate.
A nitride-based semiconductor device according to the present invention includes: a nitride-based semiconductor multilayer structure including a p-type semiconductor region, of which the surface is an m-plane; and an electrode that is arranged on the p-type semiconductor region. The p-type semiconductor region is made of an AlxInyGazN semiconductor (where x+y+z=1, x≧0, y≧0 and z≧0), and the electrode includes an Mg layer that is in contact with the surface of the p-type semiconductor region.
In one preferred embodiment, the electrode includes the Mg layer and a metal layer that has been stacked on the Mg layer, and the metal layer is made of at least one metal selected from the group consisting of Pt, Mo and Pd.
In this particular preferred embodiment, a metal alloy layer, including Mg and the at least one metal selected from the group consisting of Pt, Mo and Pd, is sandwiched between the Mg layer and the metal layer.
In another preferred embodiment, the electrode includes the Mg layer and a metal alloy layer that has been formed on the Mg layer. The metal alloy layer is made of an alloy of Mg and at least one metal selected from the group consisting of Pt, Mo and Pd.
In still another preferred embodiment, the Mg layer is made up of a film that is present on the surface of the p-type semiconductor region.
In yet another preferred embodiment, the Mg layer is made up of islands of Mg portions that are dispersed on the surface of the p-type semiconductor region.
In yet another preferred embodiment, the semiconductor multilayer structure has an active layer including an AlaInbGacN layer (where a+b+c==1, a≧0, b≧0 and c≧0) and emitting light.
In yet another preferred embodiment, the Mg layer has a thickness of 2 nm to 45 nm.
In a specific preferred embodiment, the Mg layer has a thickness of 2 nm to 15 nm.
In yet another preferred embodiment, the thickness of the Mg layer is equal to or smaller than that of the metal layer.
In yet another preferred embodiment, the concentration of Ga in the Mg layer is higher than that of nitrogen in the same Mg layer.
In a specific preferred embodiment, the concentration of Ga is ten times or more as high as that of nitrogen.
In yet another preferred embodiment, the nitride-based semiconductor device further includes a semiconductor substrate that supports the semiconductor multilayer structure.
In yet another preferred embodiment, the p-type semiconductor region is made of GaN.
A light source according to the present invention includes a nitride-based semiconductor light-emitting device, and a wavelength converter including a phosphor that changes the wavelength of the light that has been emitted from the nitride-based semiconductor light-emitting device. The nitride-based semiconductor light-emitting device includes: a nitride-based semiconductor multilayer structure including a p-type semiconductor region, of which the surface is an m-plane; and an electrode that is arranged on the p-type semiconductor region. The p-type semiconductor region is made of an AlxInyGazN semiconductor (where x+y+z==1, x≧0, y≧0 and z≧0). And the electrode includes an Mg layer that is in contact with the surface of the p-type semiconductor region.
In one preferred embodiment, the p-type semiconductor region is made of GaN.
A method for fabricating a nitride-based semiconductor device according to the present invention includes the steps of: (a) providing a substrate; (b) forming a nitride-based semiconductor multilayer structure, including a p-type semiconductor region, on the substrate, the p-type semiconductor region having an m-plane as its surface and being made of an AlxInyGazN semiconductor (where x+y+z=1, x≧0, y≧0 and z≧0); and (c) forming an electrode on the surface of the p-type semiconductor region of the semiconductor multilayer structure. The step (c) includes the step of forming an Mg layer on the surface of the p-type semiconductor region.
In one preferred embodiment, the step (c) includes the step of forming a metal layer, which is made of at least one metal selected from the group consisting of Pt, Mo and Pd, after the Mg layer has been formed.
In this particular preferred embodiment, the step (c) includes the step of subjecting the Mg layer to a heat treatment after the metal layer has been formed.
In a specific preferred embodiment, the heat treatment is carried out at a temperature of 500° C. to 700° C.
More preferably, the heat treatment is carried out at a temperature of 550° C. to 650° C.
In yet another preferred embodiment, the step of forming the Mg layer includes depositing Mg by evaporation onto the surface of the p-type semiconductor region by irradiating the surface with pulses of an electron beam.
In yet another preferred embodiment, the method includes the step of adjusting the thickness of the Mg layer to the range of 2 nm to 45 nm after the heat treatment.
In yet another preferred embodiment, the step (b) includes the step of forming a layer of an AlxInyGazN semiconductor (where x+y+z==1, x≧0, y≧0 and z≧0).
In yet another preferred embodiment, the method includes the step of removing the substrate after the step (b) has been performed.
In yet another preferred embodiment, the concentration of Ga in the Mg layer is higher than that of nitrogen in the same Mg layer.
In yet another preferred embodiment, the concentration of Ga is ten times or more as high as that of nitrogen.
In yet another preferred embodiment, the p-type semiconductor region is made of GaN.
In a method of forming an Mg thin film to be an electrode for a semiconductor device according to the present invention, the Mg thin film is formed by irradiating Mg with pulses of an electron beam and evaporating Mg.
In one preferred embodiment, the Mg thin film is formed without heating a supporting member on which the Mg thin film is going to be formed.
In a nitride-based semiconductor device according to the present invention, an electrode on a semiconductor multilayer structure includes an Mg layer that is in contact with the surface (which is an m-plane) of a p-type semiconductor region. As a result, the contact resistance can be reduced.
a) is a schematic cross-sectional view illustrating a nitride-based semiconductor light-emitting device 100 as a preferred embodiment of the present invention, and
a) through 4(c) are graphs showing how the specific contact resistance (Q·cm2) changed with the work function (eV) of a metal that was in contact with GaN.
a) is a graph showing how the specific contact resistance changed with the thickness of a thermally treated Mg layer in an Mg/Pt electrode and
a) through 6(c) are photographs showing the surface states of a thermally treated electrode in situations where the Mg layer had thicknesses of 2 nm, 15 nm and 45 nm, respectively.
a) and 10(b) are graphs showing the photoluminescence intensities of a GaN layer that was thermally treated at 800° C. and 900° C., respectively.
a) and 11(b) are profiles of Mg that were obtained in the depth direction of an (Mg/Pt) electrode structure by SIMS analysis.
a) and 12(b) are profiles of Ga that were obtained in the depth direction of the (Mg/Pt) electrode structure by SIMS analysis.
a) and 13(b) are profiles of N that were obtained in the depth direction of the (Mg/Pt) electrode structure by SIMS analysis.
a) and 14(b) show cross-sectional transmission electron microscope (TEM) photographs of an (Mg/Pt) electrode structure in which an Mg layer was deposited on an m-plane GaN layer as substitutes for drawings.
a) and 15(b) are profiles of Pt that were obtained in the depth direction of the (Mg/Pt) electrode structure by SIMS analysis.
a) illustrates a cross-sectional structure of an Mg/Pt electrode (in which an Mg layer was deposited on an m-plane GaN layer) yet to be subjected to a heat treatment, and
a) is a graph showing the current-voltage characteristic of a light-emitting diode that uses an electrode consisting of Mg/Pt layers and
a) is a graph showing the specific contact resistances of an electrode consisting of an Au layer and an electrode consisting of Mg/Au layers and
a) and 19(b) are graphs showing the results of hardness mapping that was carried out on a c-plane GaN substrate and an m-plane GaN substrate, respectively.
Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, any elements shown in multiple drawings and having substantially the same function will be identified by the same reference numeral for the sake of simplicity. It should be noted, however, that the present invention is in no way limited to the specific preferred embodiments to be described below.
a) schematically illustrates the cross-sectional structure of a nitride-based semiconductor light-emitting device 100 as a preferred embodiment of the present invention. What is illustrated in
The nitride-based semiconductor light-emitting device 100 of this preferred embodiment includes a GaN-based substrate 10, of which the principal surface 12 is an m-plane, a semiconductor multilayer structure 20 that has been formed on the GaN-based substrate 10, and an electrode 30 arranged on the semiconductor multilayer structure 20. In this preferred embodiment, the semiconductor multilayer structure 20 is an m-plane semiconductor multilayer structure that has been formed through an m-plane crystal growth and its principal surface is an m-plane. It should be noted, however, that a-plane GaN could grow on an r-plane sapphire substrate in some instances. That is why according to the growth conditions, the principal surface of the GaN-based substrate 10 does not always have to be an m-plane. In the semiconductor multilayer structure 20 of the present invention, at least the surface of its p-type semiconductor region that is in contact with an electrode needs to be an m-plane.
The nitride-based semiconductor light-emitting device 100 of this preferred embodiment includes the GaN-based substrate 10 to support the semiconductor multilayer structure 20. However, the device 100 may have any other substrate instead of the GaN-based substrate 10 and could also be used without the substrate.
b) schematically illustrates the crystal structure of a nitride-based semiconductor, of which the principal surface is an m-plane, as viewed on a cross section thereof that intersects with the principal surface of the substrate at right angles. Since Ga atoms and nitrogen atoms are present on the same atomic-plane that is parallel to the m-plane, no electrical polarization will be produced perpendicularly to the m-plane. That is to say, the m-plane is a non-polar plane and no piezoelecrtric field will be produced in an active layer that grows perpendicularly to the m-plane. It should be noted that In and Al atoms that have been added will be located at Ga sites and will replace the Ga atoms. Even if at least some of the Ga atoms are replaced with those In or Al atoms, no electrical polarization will still be produced perpendicularly to the m-plane.
Such a GaN-based substrate, of which the principal surface is an m-plane, will be referred to herein as an “m-plane GaN-based substrate”. To obtain an m-plane nitride-based semiconductor multilayer structure that has grown perpendicularly to the m-plane, typically such an m-plane GaN-based substrate may be used and semiconductors may be grown on the m-plane of that substrate. This is because the crystallographic-plane orientation of the semiconductor multilayer structure should be identical with that of the principal surface of the GaN-based substrate. However, the principal surface of the substrate does not always have to be an m-plane as described above, and the device as a final product could already have its substrate removed.
The crystal structure of a nitride-based semiconductor, of which the principal surface is a c-plane, as viewed on a cross section thereof that intersects with the principal surface of the substrate at right angles is illustrated schematically in
A c-plane GaN-based substrate is generally used to grow GaN-based semiconductor crystals thereon. In such a substrate, a Ga atom layer and a nitrogen atom layer that extend parallel to the c-plane are slightly misaligned from each other in the c-axis direction, and therefore, electrical polarization will be produced in the c-axis direction.
Referring to
The semiconductor multilayer structure 20 of this preferred embodiment has other layers, one of which is an AluGavInwN layer (where u+v+w=1, u≧0, v≧0 and w≧0) 22 that is arranged between the active layer 24 and the substrate 10. The AluGavInwN layer 22 of this preferred embodiment has first conductivity type, which may be n-type, for example. Optionally, an undoped GaN layer could be inserted between the active layer 24 and the AldGaeN layer 26.
In the AldGaeN layer 26, the mole fraction d of Al does not have to be uniform, but could vary either continuously or stepwise, in the thickness direction. In other words, the AldGaeN layer 26 could have a multilayer structure in which a number of layers with mutually different Al mole fractions d are stacked one upon the other, or could have its dopant concentration varied in the thickness direction. To reduce the contact resistance, the uppermost portion of the AldGaeN layer 26 (i.e., the upper surface region of the semiconductor multilayer structure 20) is preferably a layer that has an Al mole fraction d of zero (i.e., a GaN layer). In that case, the Mg layer 32 to be described later will be in contact with the GaN layer. However, the Al mole fraction d does not have to be zero. Alternatively, an Al0.05Ga0.95N layer, of which the Al mole fraction d is approximately 0.05, could also be used. Then, the Mg layer 32 to be described later will be in contact with that Al0.05G0.95N layer.
An electrode 30 has been formed on the semiconductor multilayer structure 20. The electrode 30 of this preferred embodiment includes an Mg layer 32 consisting essentially of Mg, on which stacked is a Pt layer consisting essentially of Pt. In the electrode 30, the Mg layer 32 is in contact with the p-type semiconductor region of the semiconductor multilayer structure 20 and functions as a portion of a p-(or p-side) electrode. In this preferred embodiment, the Mg layer 32 is in contact with the AldGaeN layer 26 that is doped with a dopant of a second conductivity type (e.g., p-type), which may be Mg. Examples of other preferred p-type dopants include Zn and Be.
The metal layer 34 that is in contact with the surface of the Mg layer 32 does not have to be a Pt layer but may also be a layer of a metal that would make an alloy with Mg less easily than Au. Stated otherwise, the metal layer 34 to be in contact with the Mg layer 32 should not be made of Au (gold) that would make an alloy with Mg easily. The Mg layer 32 does not make an alloy with Pt or any other suitable metal that constitutes the metal layer 34. It should be noted, however, that “does not make an alloy with Pt or any other suitable metal” could also refer to a situation where Mg includes a metal such as Pt at a concentration of less than a few percent (which may be less than one percent). In other words, “make an alloy with a metal such as Pt” refers to a situation where Mg includes a metal such as Pt at a concentration of at least a few percent (which may be equal to or greater than one percent). The Mg layer 32 and the metal layer 34 could also contain some impurities to be inevitably included during the process steps of forming those layers.
Optionally, there could be a metal alloy layer containing Mg between the Mg layer 32 and the metal layer 34. Pt, Mo and Pd are metals that would make an alloy with Mg less easily than Au. But a metal alloy layer could still be produced through a reaction to a portion of the Mg layer 32 as a result of the heat treatment to be described later.
If a relatively thin metal layer has been deposited on the Mg layer 32, then all of that thin metal layer could make an alloy with some of Mg atoms in the Mg layer. In that case, there would be only that metal alloy layer on the Mg layer.
In any case, on the electrode with any of various combinations of metals described above, not just the metal layer 34 or the metal alloy layer but also an electrode layer or an interconnect layer made of a different metal or metal alloy could be stacked.
In this preferred embodiment, the electrode 30 may have a thickness of 10 nm to 200 nm, for example. In the electrode 30, the Mg layer 32 may have a thickness of 2 nm to 45 nm. It should be noted that in this case, the thickness of the Mg layer 32 is supposed to be that of the Mg layer that has already been thermally treated.
On the other hand, the metal layer 34, which is made of at least one metal selected from the group consisting of Pt, Mo and Pd and which has been stacked on the Mg layer 32, may have a thickness of 200 nm or less (or fall within the range of 10 nm to 200 nm). It should be noted that the Mg layer 32 is preferably thinner than the metal layer 34. This is because if the strain balance were lost between the Mg layer 32 and the metal layer 34, the Mg layer 32 could peel off from the AldGaeN layer 26.
Meanwhile, the GaN-based substrate 10, of which the principal surface 12 is an m-plane, may have a thickness of 100 μm to 400 μm, for example. This is because if the wafer has a thickness of at least approximately 100 μm, then there will be no trouble handling such a wafer. It should be noted that as long as the substrate 10 of this preferred embodiment has an m-plane principal surface 12 made of a GaN-based material, the substrate 10 could have a multilayer structure. That is to say, the GaN-based substrate 10 of this preferred embodiment could also refer to a substrate, at least the principal surface 12 of which is an m-plane. That is why the entire substrate could be made of a GaN-based material. Or the substrate may also be made of the GaN-based material and another material in any combination.
In the structure of this preferred embodiment, an electrode 40 has been formed as an n-side electrode on a portion of an n-type AluGavInwN layer 22 (with a thickness of 0.2 μm to 2 μm, for example) on the substrate 10. In the example illustrated in
In this preferred embodiment, the active layer 24 has a GaInN/GaN multi-quantum well (MQW) structure (with a thickness of 81 nm, for example) in which Ga0.9In0.1N well layers (each having a thickness of 9 nm, for example) and GaN barrier layers (each having a thickness of 9 nm, for example) are alternately stacked one upon the other.
On the active layer 24, stacked is the p-type AldGaeN layer 26, which may have a thickness of 0.2 μm to 2 μm. Optionally, an undoped GaN layer could be inserted between the active layer 24 and the AldGaeN layer 26 as described above.
In addition, a GaN layer of the second conductivity type (which may be p-type, for example) could be formed on the AldGaeN layer 26. Furthermore, a contact layer of p+-GaN and the Mg layer 32 could be stacked in this order on that GaN layer. In that case, the GaN contact layer could also be regarded as forming part of the AldGaeN layer 26, not a layer that has been stacked separately from the AldGaeN layer 26.
Hereinafter, the features of this preferred embodiment will be described in further detail with reference to
First of all,
Generally speaking, the contact resistance is inversely proportional to the area S (cm2) of the contact. In this case, supposing R (Ω) is contact resistance, the equation R=Rc/S should be satisfied. The constant of proportionality Rc is called “specific contact resistance” and corresponds to contact resistance R when the contact area S is 1 cm2. That is to say, the magnitude of the specific contact resistance does not depend on the contact area S and can be used as an index evaluating the characteristics of a contact. In the following description, the “specific contact resistance” will sometimes be referred to herein as just “contact resistance”.
Specifically,
As can be seen from
On the other hand, as for results that were obtained by thermally treating the metal deposited, it can be seen from
a) is a graph showing the relationship between the specific contact resistance and the thickness of an Mg layer in an Mg/Pt electrode (in which a Pt layer was stacked on the Mg layer). In this case, the thickness of the Pt layer yet to be thermally treated was fixed at 75 nm. On the other hand,
The data shown in
The Mg/Pt electrode and the Pd/Pt electrode are both in contact with the Mg-doped m-plane GaN layer. The m-plane GaN layer that is in contact with these electrodes is doped with 7×1019 cm−3 Mg to a depth of 20 nm as measured from the surface (i.e., the uppermost surface region with a thickness of 20 nm). On the other hand, the rest of the m-plane GaN layer, of which the depth exceeds 20 nm as measured from the surface, is doped with 1×1019 cm−3 Mg. If the concentration of the p-type dopant is locally increased in this manner in the uppermost surface region of the GaN layer that is in contact with the p-electrode, the contact resistance can be reduced to the lowest possible level. On top of that, by adopting such a doping scheme, the in-plane non-uniformity of the current-voltage characteristic can also be reduced. As a result, the variation in drive voltage between respective chips can also be reduced. That is why in every experimental example disclosed in this application, the surface region of the p-type GaN layer that is in contact with the electrode is doped with 7×1019 cm−3 Mg to a depth of 20 nm as measured from the surface, while the other deeper region is doped with 1×1019 cm−3 Mg. It should be noted that the contact resistance of Mg shown in
In the graph shown in
The graph of
The results of experiments the present inventors carried out revealed that once the thickness of the Mg layer exceeded 45 nm, the contact resistance of the Mg/Pt electrode with respect to the m-plane GaN layer was approximately equal to that of the Pd/Pt electrode with respect to the m-plane GaN layer (shown in
The present inventors confirmed that in a range where the Mg layer had a thickness of 45 nm or less, the thinner the Mg layer, the lower the contact resistance. When the Mg layer thickness decreased to the vicinity of 15 nm, the contact resistance started to drop steeply with the decrease in the thickness of the Mg layer. And when the Mg layer thickness was around 2 nm, the lowest possible contact resistance was achieved.
In view of these considerations, in the semiconductor device to be obtained as a final product by going through every manufacturing process step including the heat treatment, the Mg layer 32 preferably has a thickness of 45 nm or less, and more preferably falls within the range of 2 nm to 15 nm.
a) through 6(c) are photographs showing the surface states of a thermally treated electrode in situations where the Mg layer had thicknesses of 2 nm, 15 nm and 45 nm, respectively. In this case, the thickness of the Mg layer was measured after having been thermally treated at 600° C. for 10 minutes.
As shown in
It should be noted that if the Mg layer had a thickness of approximately 15 nm or less, the surface of the electrode had an extremely high degree of planarity. That is why the Mg layer more preferably has a thickness of 15 nm or less.
Before being subjected to the heat treatment, the respective layers had thicknesses shown in the following Table 1:
Their heat treatment temperatures and process times are as shown in the following Table 2:
As is clear from
Next, it will be described with reference to
Before being subjected to the heat treatment, the respective layers had the thicknesses shown in the following Table 3:
First of all, in the electrode consisting of Pd/Pt layers, the m-plane GaN had higher contact resistance than the c-plane GaN. Also, in both of the m- and c-plane GaN, the contact resistance increased at a heat treatment temperature exceeding 500° C.
On the other hand, before subjected to the heat treatment, the electrode of the m-plane GaN (Mg/Pt) had higher contact resistance than the electrode consisting of the Pd/Pt layers. This result conforms to a technical commonplace that the smaller the work function of a metal is, the higher its contact resistance will be. However, in the electrode of the m-plane GaN (Mg/Pt), the higher the heat treatment temperature, the lower the contact resistance. And at a heat treatment temperature of 500° C., the contact resistance of the m-plane GaN (Mg/Pt) was equal to or lower than that of the m-plane GaN (Pd/Pt).
Besides, at a temperature exceeding 500° C. (e.g., at 600° C.), the contact resistance of the m-plane GaN (Mg/Pt) further decreased to get equal to, or even lower than, that of the c-plane GaN (Mg/Pt). According to the graph shown in
At a heat treatment temperature of 600° C., the contact resistance of the m-plane GaN (Mg/Pt) was lower than those of the m- and c-plane GaN (Mg/Pt) at 600° C. and more specifically decreased to approximately 1.0E−02 Ωcm−2 or the vicinity thereof. Furthermore, at a heat treatment temperature of 700° C., the contact resistance of the m-plane GaN (Mg/Pt) had increased from its resistance at 600° C. but was still lower than the contact resistances of the m- and c-plane GaN (Mg/Pt) at 700° C.
Consequently, the heat treatment temperature of the m-plane GaN (Mg/Pt) is preferably equal to or higher than 500° C., for example. However, if the heat treatment temperature exceeded 700° C. to reach a predetermined temperature of 800° C., for example, then the film quality of the electrode and the GaN layer would start to deteriorate noticeably. That is why the upper limit is preferably set to be 700° C. or less, and the heat treatment temperature is more preferably in the vicinity of 600° C. (e.g., 600° C.±50° C.).
As can be seen from
On the other hand, if Pd and Pt layers were stacked in this order on a p-type m-plane GaN layer (i.e., in the case of M-GaN (Pd/Pt)), some degree of deterioration such as surface roughening was seen on the surface of the metal at heat treatment temperatures of 600° C. and 700° C. According to the results of AFM measurements, the surface had an Ra of about 30 nm and an Ra of about 77 nm at 600° C. and 700° C., respectively. That is to say, it can be seen that such deterioration of the electrode that has gone through a heat treatment is a problem peculiar to the m-plane GaN electrode.
However, if Mg and Pt layers were stacked in this order on a p-type m-plane GaN layer (i.e., in the case of M-GaN (Mg/Pt)), the present inventors confirmed that a very small degree of unevenness was seen at the heat treatment temperature of 700° C. but that the electrode never deteriorated at any of these heat treatment temperatures of 500° C., 600° C. and 700° C. According to the results of AFM measurements, the surface had Ra of about 1.5 nm, about 1.5 nm and about 4.5 nm at 500° C., 600° C. and 700° C., respectively. Thus, good surface states were achieved. And in the structure of this preferred embodiment, the electrode surface preferably has an Ra of about 4.5 nm or less, more preferably about 1.5 nm or less.
The present inventors deposited an Mg layer to a thickness of 30 nm on a GaN layer, thermally treated it at 800° C. and at 900° C. for 10 minutes, and then measured the photoluminescence of the GaN layer. The results are shown in
First of all, the present inventors discovered and confirmed via experiments that if the heat treatment was carried out at 700° C. or less, the photoluminescence spectrum after the heat treatment remained virtually the same as the one before the heat treatment. On the other hand, if the heat treatment was carried out at 800° C., a photoluminescence called “yellow band”, which would have been caused due to the presence of vacancy defects, appeared around 530 nm as shown in
In any of these electrodes, before the heat treatment, the Mg layer had a thickness of 7 nm and the Pt layer had a thickness of 75 nm.
In the graphs shown in
In the graphs shown in
As shown in
If the heat treatment was carried out, Mg on the c-plane GaN was seen to have diffused into the p-type GaN layer at rather high concentrations as shown in
a) shows the results that were obtained in a situation where an Mg/Pt electrode, in which an Mg layer was stacked on the GaN layer, was not subjected to heat treatment (i.e., in the as-deposited state). On the other hand,
As shown in
Specifically, the present inventors confirmed that if the heat treatment was carried out, Ga diffused into the Mg layer as shown in
More specifically, in the case of the c-plane GaN, Ga diffused into both the Mg and Pt layers and moved into the electrode even from deep inside GaN crystals. In other words, in the c-plane GaN, Ga atoms diffused significantly almost everywhere from the GaN layer into the electrode. In the m-plane GaN, on the other hand, when the heat treatment temperature was 600° C., it seems that Ga atoms moved only in the vicinity of the interface unlike the c-plane GaN. Compared to a c-plane, it would be less easy for the atoms to move freely from an m-plane. Even in the m-plane GaN, however, if the heat treatment temperature was 630° C., Ga atoms diffused into the entire Mg and Pt layers. It should be noted that the m-plane contact resistance was lower at a heat treatment temperature of 600° C. than at 630° C. This is probably because at a heat treatment temperature of 600° C., it would be hard for nitrogen atoms to diffuse from the m-plane as will be described later, and therefore, Ga vacancies would act as acceptors as a result. On the other hand, at a heat treatment temperature of 630° C., a greater number of nitrogen atoms would diffuse toward the electrode than at 600° C.
a) is a graph showing the profile of nitrogen atoms in the depth direction in an Mg/Pt electrode yet to be subjected to a heat treatment, while
It can be seen from
On the other hand, it was confirmed that in the electrode on the c-plane GaN that had been thermally treated, nitrogen atoms diffused toward the electrode as shown in
Next, the relation between N (nitrogen) and Ga concentrations in an Mg layer will be described in detail. As shown in
Furthermore, no matter whether the heat treatment was carried out at 600° C. or 630° C., the concentration of N (nitrogen) atoms that was detected in the Mg layer on the m-plane GaN with an SIMS was 1×1.0 E−03, which is the limit of responsivity, as shown in
On the other hand, in the Mg layer on the c-plane GaN layer, both Ga and N had an SIMS detected intensity of 1E−2, and therefore, would have been contained there in a concentration of approximately 1×1019 cm−3. That is to say, in the Mg layer on the c-plane GaN layer, the Ga atoms and N atoms had approximately the same concentration.
These elements (namely, Mg, Ga, N and Pt) would behave in a similar manner even if part of Ga is replaced with Al or In in the GaN layer that is in contact with the Mg layer. Their behavior would also be the same even if the GaN-based semiconductor layer that is in contact with the Mg layer is doped with a non-Mg element as a dopant.
Next, look at
In this example, the Mg layer was deposited to a thickness of 7 nm on GaN crystals as shown in
As can be seen from
As shown in
These results reveal that according to the structure (Mg/Pt) of this preferred embodiment, in a region where Mg was in contact with the GaN layer, their alloy was produced much less than in the case of Au.
The present inventors made a sample in which an Mg layer was deposited to a thickness of 2 nm on an m-plane GaN and then thermally treated at 600° C. for 10 minutes. And we observed cross sections of that sample with a transmission electron microscope (TEM).
If the Mg layer as deposited was thicker than 5 nm, the thickness of the Mg decreased after having been thermally treated at 600° C. for 10 minutes. However, the Mg layer was still present as a substantially continuous film even after the heat treatment. On the other hand, it was confirmed that if the Mg layer as deposited had a thickness of approximately 2 nm, sometimes there were islands of Mg, which had not been absorbed into an Mg—Pt alloy, after the heat treatment had been carried out at 600° C. for 10 minutes as shown in
It should be noted that the “Mg layer” could also refer herein to a group of such a huge number of islands of Mg on the surface of a p-type semiconductor region. Alternatively, the “Mg layer” could even be a film with a lot of openings (such as a porous film). In this manner, as long as Mg that has not been corroded by Pt contacts with the surface of a p-type semiconductor region (that is an m-plane), the contact resistance can be reduced sufficiently effectively.
Also, even if an Mo layer or a Pd layer is deposited in place of the Pt layer on the Mg layer, the contact resistance could also be reduced substantially no less effectively. To achieve that effect of reducing the contact resistance, the point is that an Mg layer and a GaN-based semiconductor should contact with each other and that a metal on the Mg layer should be a material that would produce an alloy with Mg less easily than Au.
Hereinafter, the configuration of this preferred embodiment will be described in further detail with reference to
As shown in
In the semiconductor multilayer structure 20, an active layer 24 including an AlaInbGacN layer (where a+b+c=1, a≧0, b≧0 and c≧0) has been formed on the AlxGayInzN layer 22. The active layer 24 consists of InGaN well layers with an In mole fraction of approximately 25% and GaN barrier layers, both the well layers and the barrier layers may have a thickness of 9 nm each, and the well layers may have a well layer period of three. On the active layer 24, stacked is an AldGaeN layer (where d+e=1, d≧0 and e≧0) 26 of the second conductivity type (which may be p-type, for example), which may be an AlGaN layer with an Al mole fraction of 10% and may have a thickness of 0.2 μm. In this preferred embodiment, the AldGaeN layer 26 is doped with Mg as a p-type dopant to a level of approximately 1018 cm−3, for example. Also, in this example, an undoped GaN layer (not shown) is interposed between the active layer 24 and the AldGaeN layer 26.
Furthermore, in this example, on the AldGaeN layer 26, stacked is a GaN layer (not shown) of the second conductivity type (which may be p-type, for example). In addition, on the contact layer of p+-GaN, stacked in this order are an Mg layer 32 and a Pt layer 34. And this stack of the Mg layer 32 and the Pt layer 34 is used as an electrode (i.e., a p-electrode) 30.
This semiconductor multilayer structure 20 further has a recess 42 that exposes the surface of the AluGavInwN layer 22. And an electrode 40 (n-electrode) has been formed on the AluGavInwN layer 22 at the bottom of the recess 42, which may have a width (or diameter) of 20 μm and a depth of 1 μm, for example. The electrode 40 may have a multilayer structure consisting of Ti, Al and Pt layers, which may have thicknesses of 5 nm, 100 nm and 10 nm, respectively.
The present inventors discovered that the nitride-based semiconductor light-emitting device 100 of this preferred embodiment could have an operating voltage Vop that was approximately 1.5 V lower than that of a conventional m-plane LED with a Pd/Pt electrode, and therefore, could cut down the power dissipation as a result.
Hereinafter, it will be described with reference to
First of all, an m-plane substrate 10 is prepared. In this preferred embodiment, a GaN substrate is used as the substrate 10. The GaN substrate of this preferred embodiment is obtained by HVPE (hydride vapor phase epitaxy).
For example, a thick GaN film is grown to a thickness of several nanometers on a c-plane sapphire substrate, and then diced perpendicularly to the c-plane (i.e., parallel to the m-plane), thereby obtaining m-plane GaN substrates. However, the GaN substrate does not have to be prepared by this particular method. Alternatively, an ingot of bulk GaN may be made by a liquid phase growth process such as a sodium flux process or a melt-growth method such as an ammonothermal process and then diced parallel to the m-plane.
The substrate 10 does not have to be a GaN substrate but may also be a gallium oxide substrate, an SiC substrate, an Si substrate or a sapphire substrate, for example. To grow an m-plane GaN-based semiconductor on the substrate by epitaxy, the principal surface of the SiC or sapphire substrate is preferably also an m-plane. However, in some instances, a-plane GaN could grow on an r-plane sapphire substrate. That is why according to the growth conditions, the surface on which the crystal growth should take place does not always have to be an m-plane. In any case, at least the surface of the semiconductor multilayer structure 20 should be an m-plane. In this preferred embodiment, crystal layers are formed one after another on the substrate 10 by MOCVD (metalorganic chemical vapor deposition) process.
Next, an AluGavInwN layer 22 is formed on the m-plane GaN substrate 10. As the AluGavInwN layer 22, AlGaN may be deposited to a thickness of 3 μm, for example. A GaN layer may be deposited by supplying TMG(Ga(CH3)3), TMA(Al(CH3)3) and NH3 gases onto the m-plane GaN substrate 10 at 1,100° C., for example.
Subsequently, an active layer 24 is formed on the AluGavInwN layer 22. In this example, the active layer 24 has a GaInN/GaN multi-quantum well (MQW) structure in which Ga0.9In0.1N well layers and GaN barrier layers, each having a thickness of 9 nm, have been stacked alternately to have an overall thickness of 81 nm. When the Ga0.9In0.1N well layers are formed, the growth temperature is preferably lowered to 800° C. to introduced In.
Thereafter, an undoped GaN layer is deposited to a thickness of 30 nm, for example, on the active layer 24, and then an AldGaeN layer 26 is formed on the undoped GaN layer. As the AldGaeN layer 26, p-Al0.14Ga0.86N is deposited to a thickness of 70 nm by supplying TMG, NH3, TMA, TMI gases and Cp2Mg (cyclopentadienyl magnesium) gas as a p-type dopant.
Next, a p-GaN contact layer is deposited to a thickness of 0.5 μm, for example, on the AldGaeN layer 26. In forming the p-GaN contact layer, Cp2Mg is supplied as a p-type dopant.
Thereafter, respective portions of the p-GaN contact layer, the AldGaeN layer 26, the undoped GaN layer, and the active layer 24 are removed by performing a chlorine-based dry etching process, thereby making a recess 42 and exposing a region of the AlxGayInzN layer 22 where an n-electrode will be formed. Then, Ti/Pt layers are deposited as an n-electrode 40 on the region reserved for an n-type electrode at the bottom of the recess 42.
Subsequently, an Mg layer 32 is formed on the p-GaN contact layer and then a Pt layer 34 is stacked on the Mg layer 32, thereby forming a p-electrode 40. In this preferred embodiment, a technique for performing an evaporation process while evaporating the material metal in pulses (i.e., a so-called “pulse evaporation process”) is used to form the Mg layer 32. More specifically, metal Mg contained in a crucible in a vacuum (of 5×10−6 Torr, for example) is irradiated with pulses of electron beam, thereby evaporating the material metal in pulses. Some of the molecules or atoms of that material metal are deposited on the p-GaN contact layer, thereby forming an Mg layer 32. Those pulses may have a pulse width of 0.5 seconds and may be applied repeatedly at a frequency of 1 Hz. The pulse width preferably falls within the range of 0.005 seconds through 5 seconds and the pulse frequency preferably falls within the range of 0.1 Hz through 100 Hz. By adopting such a technique, a dense film of quality could be formed as the Mg layer 32. The Mg layer had such high density probably because by performing such a pulse evaporation, Mg atoms or a cluster of Mg atoms that collided against the p-GaN contact layer would have their kinetic energy increased. That is to say, when irradiated with an electron beam, a portion of the material Mg instantaneously turn into high-energy Mg atoms and vaporize or evaporate. Then, those Mg atoms reach the p-GaN contact layer. On reaching the p-GaN contact layer, the Mg atoms make migration, thereby forming a dense and homogeneous Mg thin film at an atomic level. By applying a pulse of an electron beam, an Mg thin film consisting of approximately one to twenty atomic layers is formed. And by applying multiple pulses of electron beams a number of times, those Mg thin films will be stacked one upon the other on the p-GaN contact layer, thereby forming an Mg layer 32 to a desired thickness. The electron beam preferably has a peak intensity that is high enough to supply the Mg atoms such kinetic energy that would cause those Mg atoms adsorbed to migrate easily. Also, the driving power of the electron gun is preferably determined so that the Mg thin film is deposited to a thickness of 20 atomic layers (corresponding to approximately 5 nm) or less per pulse of an electron beam. This is because if the Mg thin film deposited per pulse of the electron beam were thicker than 20 atomic layers, then it would be difficult to obtain a dense and homogenous Mg thin film. The deposition rate is more preferably five atomic layers or less per pulse of the electron beam. The reason is that if there were too many Mg atoms, those Mg atoms would collide against each other while migrating and would lose their kinetic energy.
Generally speaking, Mg is an element that gets oxidized easily when exposed to water or the air. When put in the air, an Mg thin film that has been deposited on a supporting substrate by a normal evaporation process will get oxidized quickly. As a result, the Mg thin film will gradually lose its metallic gloss and will eventually get crumbly and peel off from its supporting member. On the other hand, the Mg layer 32 that has been formed by the method of this preferred embodiment (i.e., the pulse evaporation process) is highly dense and homogenous at an atomic level and has a structure, of which the atomic ordering is so aligned that it looks as if it had been grown epitaxially. On top of that, since it has almost no pinholes that would cause the oxidation, the layer will get hardly oxidized. That is why even if left in the air for a few months, the layer can still maintain a shiny mirror surface.
Also, in general, Mg will vaporize instantaneously by nature once its temperature has exceeded its melting point. For that reason, it is very difficult to precisely control the thickness of an Mg thin film to deposit on the order of nanometers by adjusting the deposition rate of the Mg thin film. On the other hand, the Mg layer 32 that has been formed by the method of this preferred embodiment can have its thickness controlled very precisely on the order of nanometers by appropriately setting the duration of a pulsed electron beam radiated.
It should be noted that the method of this preferred embodiment could also be used effectively to deposit an Mg thin film on an Si substrate or a glass substrate. That is why the method of this preferred embodiment can be used to form an Mg thin film on not just GaN but also any of various other substrates. And the method of this preferred embodiment can be used effectively to form an electrode in the field of semiconductor processing, among other things. Also, according to the method of this preferred embodiment, there is no need to heat either the substrate or the supporting member while an Mg thin film is being deposited by evaporation. For that reason, a highly dense and homogenous Mg thin film can be deposited at around room temperature even on a substrate or supporting member that has been formed in a shape that makes it difficult to heat it.
Furthermore, the Mg layer 32 that has been formed by the method of this preferred embodiment can still maintain good stability even when thermally treated at a temperature of 600° C. or more.
This preferred embodiment uses a technique for depositing an Mg layer 32 while evaporating the material metal (i.e., metal Mg) in pulses. However, as long as the Mg layer 32 can be formed, any other technique can also be adopted. As an alternative method for forming such a dense Mg layer of quality, a thermal CVD process or a molecular beam epitaxy (MBE) could also be used.
Optionally, the substrate 10 and a portion of the AluGavInwN layer 22 could be removed after that by some technique such as laser lift-off, etching or polishing. In that case, either only the substrate 10 or the substrate 10 and a portion of the AluGavInwN layer 22 could be removed selectively. It is naturally possible to leave the substrate 10 and the AluGavInwN layer 22 as they are without removing them. By performing these process steps, the nitride-based semiconductor light-emitting device 100 of this preferred embodiment is completed.
In the nitride-based semiconductor light-emitting device 100 of this preferred embodiment, when a voltage is applied to between the n- and p-electrodes 40 and 30, holes are injected from the p-electrode 30 into the active layer 24 and electrons are injected from the n-electrode 40 into the active layer 24, thus producing photoluminescence with a wavelength of about 450 nm.
a) shows the current-voltage characteristic of a light-emitting diode that uses an electrode consisting of Mg/Pt layers on m-plane GaN. For your reference, the characteristics of comparative light-emitting diodes that have the same LED nitride-based semiconductor structure but that use an electrode consisting of Pd/Pt layers and an electrode consisting of Mg/Pt layers on a c-plane GaN, respectively, are also shown in
In each of these light-emitting diodes, an n-type GaN layer, an active layer in which three InGaN well layers and two GaN barrier layers are alternately stacked one upon the other, and a p-type GaN layer are stacked in this order on either an m-plane GaN substrate or a c-plane GaN substrate. In addition, on the p-type GaN layer, arranged as a p-electrode is either an Mg/Pt electrode or a Pd/Pt electrode. On the other hand, an n-electrode is formed on the n-type GaN layer by etching the p-type GaN layer and the active layer and exposing the n-type GaN layer.
As can be seen easily from
Next, a light-emitting diode that uses an electrode consisting of Mg/Pt layers on m-plane GaN (which will be referred to herein as an “m-plane light-emitting diode”) and a light-emitting diode that uses an electrode consisting of Mg/Pt layers on c-plane GaN (which will be referred to herein as a “c-plane light-emitting diode”) will be compared to each other. The m-plane light-emitting diode had a lower rise voltage than the c-plane light-emitting diode, which allowed us to confirm that the contact resistance was certainly reduced. In the m-plane light-emitting diode, for example, a current value of 20 mA could be obtained for a drive voltage of 3.2 V. As for the c-plane light-emitting diode, on the other hand, a current value of 4.8 mA was achieved for the same drive voltage. Since the optical power of a light-emitting diode will depend on the current value, it can be seen that the optical power produced by the m-plane light-emitting diode was approximately four times as much as the one produced by the c-plane light-emitting diode at a drive voltage of 3.2 V.
Also, the gradient of the curve representing the current-voltage characteristic is steeper in the device with the Mg/Pt electrode than in the device with the Pd/Pt electrode. Generally speaking, a light-emitting diode is a diode that has a p-n junction inside. And a curve representing the current-voltage characteristic of a p-n junction diode is generally approximated by the following equation:
I=I
0 exp(V/n·KT)
where I is the value of current flowing through the p-n junction diode, I0 is the current constant, V is the applied voltage, K is the Boltzmann constant, T is the temperature, n is a value representing the degree of idealness of the diode. Since the experiments were carried out at room temperature, KT=0.25 V.
The n value of a p-n junction diode is determined by the gradient of a curve representing the current-voltage characteristic. In an ideal p-n junction diode, n=1. In an actual p-n junction diode, however, the n value is different from one. But the closer to one the n value, the better. According to the results of experiments the present inventors carried out, n=1.4 in the device with the Mg/Pt electrode and n=2.2 in the device with the Pd/Pt electrode. As can be seen from these results, a good diode characteristic is realized by using the Mg/Pt electrode.
As shown in
Next, examples that used an electrode consisting of an Au layer and an electrode consisting of Mg/Au layers will be described as comparative examples with reference to
As can be seen from the results shown in
b) is a photograph representing the surface of a thermally treated electrode consisting of Mg/Au layers as a substitute for a drawing, while
Next, look at
While the present invention has been described with respect to preferred embodiments thereof, this invention is in no way limited to those specific preferred embodiments but could be modified in numerous ways and may assume many embodiments other than those specifically described above.
Even though its structure is essentially different from the preferred embodiment of the present invention, related structures are also disclosed in Patent Documents Nos. 3 and 4. However, those Patent Documents Nos. 3 and 4 do not mention at all that the crystallographic-plane of their gallium nitride-based semiconductor layer is an m-plane but just disclose a technique for forming an electrode on a c-plane gallium nitride-based semiconductor layer. More specifically, Patent Document No. 3 discloses a structure in which an Au layer is stacked on an Mg layer. And even if an electrode with such a multilayer structure were formed on an m-plane, the effect of the electrode of this preferred embodiment would never be achieved. Meanwhile, Patent Document No. 4 mentions metal layers of Ni, Cr and Mg but discloses only a specific example about an electrode structure that uses an Ni layer as the lower layer. Both of these Patent Documents Nos. 3 and 4 relate to an electrode structure that has been formed on a c-plane gallium nitride-based semiconductor layer and teach neither a problem nor a solution about the contact resistance with respect to an m-plane gallium nitride-based semiconductor layer.
The light-emitting device of the present invention described above could be used as it is as a light source. However, if the light-emitting device of the present invention is combined with a resin including a phosphor that produces wavelength conversion, for example, the device of the present invention can be used effectively as a light source with an expanded operating wavelength range (such as a white light source).
In the preferred embodiments described above, the p-type semiconductor region that is in contact with the Mg layer 32 is supposed to be made of GaN or AlGaN. However, the p-type semiconductor region could also be a layer including In such as InGaN. In that case, the contact layer to be in contact with the electrode 30 could be made of In0.2Ga0.8N with an In mole fraction of 0.2, for example. If In is included in GaN, the bandgap of AlaGabN layer (where a+b=1, a≧0 and b≧0) can be smaller than that of GaN, and therefore, the contact resistance can be reduced as a result. Consequently, the p-type semiconductor region that is in contact with the Mg layer may be made of an AlxInyGazN semiconductor (where x+y+z=1, x≧0, y≧0 and z≧0).
The effect of reducing the contact resistance can naturally be achieved by a non-LED light-emitting device (such as a semiconductor laser diode) or a device other than a light-emitting device (such as a transistor or a photodetector).
The actual m-plane does not always have to be a plane that is exactly parallel to an m-plane but may be slightly tilted from the m-plane by 0±1 degree.
According to the present invention, it is possible to reduce the contact resistance of a GaN-based semiconductor device fabricated by producing crystal growth on an m-plane substrate or a GaN-based semiconductor multilayer structure, of which the principal surface is an m-plane. As a result, although such a GaN-based semiconductor device fabricated by producing crystal growth on an m-plane substrate (or a GaN-based semiconductor multilayer structure, of which the principal surface is an m-plane) has been difficult to use extensively owing to its bad contact resistance characteristic, its industrial applicability can be expanded significantly by the present invention.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2008-285155 | Nov 2008 | JP | national |
| 2009-030147 | Feb 2009 | JP | national |
| 2009-097684 | Apr 2009 | JP | national |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 12679347 | Mar 2010 | US |
| Child | 13708136 | US |
